Wednesday, March 17, 2010

Four basic rules for news on surveys

There orders me a reader (thank you, Inés) a linkage to the fantastic page PHD Comics, which satirizes normally the academic life, but which this time has written in only one strip, four basic lessons of statistics for journalists.
"Piled Higher and Deeper" by Jorge
My rapid translation:
Dear mass media,
On having given news on surveys results, please, pay attention to the following ones sugerencias:1. If two surveys differ in less than the error margin, there is no news. (In the vignette): "The survey information for the candidate O is collapsing!!!")
2. The scientific facts do not decide for polls of public opinion (In the vignette): Network of medieval news. "Yes, Galilean, but: what does he say to us about the last surveys that they say that the ground is flat?"
3. A survey of his Internet spectators / readership is not a scientific survey (In the vignette): Announcer: And now, to refill emission time, a survey that shows that the people who thinks how I it agrees with me. (In the cartel) I am right? Of course 99 % Me divide of laugh 67 %
4. How if all the surveys include the option "does not it matter for me? (In the vignette): Here the results of the survey are votaciónCandidato A: 30%Candidato B: 26%Lo that is: 44 %
Signed: someone who gave a basic statistics course
That should sound quite so familiarly in a cómic who writes himself in the United States, it makes us see that, as a friend says, "everywhere boil beans".

No comments:

Post a Comment